Human skin detection using histogram processing and Gaussian Mixture Model based on color spaces

Skin plays important role in digital image processing. Detection of human skin color is important in numerous applications. Skin as a feature is used to form shape and geometry of the object in the image or frame. In this paper, the detection of skin region aims to find the presence of people by dividing image into skin and non-skin pixel. HSV and YCbCr color spaces are used to transform pixel and training. Histogram and Gaussian Mixture Model (GMM) is used for classification of skin and non-skin pixels in image. These models are used to describe pixels as skin distribution. Fusion approach includes grouping of several stages. Each stage has certain number of features. Different stages are used to additive feature extraction for face detection systems. Skin classifier gives threshold for skin for the given color space. The experiment is performed on three datasets namely ETHZ PASCAL dataset, Partheepan dataset and SFA dataset. The hybrid approach is used for skin detection. The accuracy, F-Score, true positive rate and false positive rate is used to measure the performance of the system.